Research Methods #2
confirmability
a degree of neutrality or the extent to which the findings of a study are shaped by the respondent and not researcher bias, motivation or interest
index
a measure in which a researcher adds or combines several distinct indicator of a construct into a single score. they are usually interval/ratio. ex. likert scale. always at the ordinal level. the number indicates your score on the scale. you're capturing an intensity of an emotion, behaviour, etc.
scale
a measure in which a researcher captures the intensity, direction, level or potency of a variable construct. they are usually ordinal.
random sample
a sample where in every element has an equal probability of selection
natural experiments
a special type of field experiment where a researcher can examine the impact of some policy or other change before and after said change (or treatment). situations that may create appropriate circumstances for a natural experiment, ex. policy changes, weather events and natural disasters
ethnography
an approach to field research emphasizing detailed descriptions of a different culture from the viewpoint of an insider who is permitted entrance/access. comes from cultural anthro & based on naturalism
concept
an idea expressed as a symbol or in words (ex poverty, racism and intelligence). how we measure concepts varies between QN & QL researchers
statistic
any piece of info you get from that sample
purposive
experts that can guide you to your sample
natural setting
field and nature experiments. more typical in social sciences.
procedure
focus group; between 6-12 people, do not want less than that (people feel like they are on the spot) but if its too big its a problem and last about 90 minutes.
the moderator
focus group; important person, their job is to keep people on track, they have the interview guide (approx. 1-3 questions meant to build the conversation among people), they present the questions and help probe them to talk, allows everyone to talk not just one person and make sure the social dynamic of the group is not in a way that is leading answers
threats to external validity
generalizability
sampling
goal of this is to get a representative collection of units or cases from a much larger collection or population, such that the research can study the smaller group and produce generalizations about the larger group
experimental group
group that receives treatment
mail and self administered surveys
hand-administered to people directly OR mailed out to households. mail surveys: mail distributions and return practices, response rates: between 50%-70%. additional factors: sponsorship & inducement to respond
descrete
have a fixed set of separate values or variable attributes (ex. gender, religion, martial status,etc)
quasi filter
here's a question, do you agree, disagree or have no opinion. ex. here if a question about another country "russian leaders are basically trying to get along with america". do you agree, disagree or have no opinion on that?
field research
learning about, understanding or describing a group of interacting people, questions: "what is the social world of X like? how do people do X in the social world?", participant observation, and members. get out there and watch and take notes. people are in natural setting that we are observing
simple questions
that use a concrete indicator ex. what is your age in years?_______
internal validity
the ability to eliminate alternative explanations of the dependent variable
generalizability
the degree to which the conclusions in your study would hold for other persons in other places and at other times.
confidence level
the estimated probability (ex. 95% or 2sd's, 99% of 2 sd's) that a given confidence interval captures the true population parameter. ex. we get a statistic of 3%, its sample, we aren't sure if it's the true population parameter, because of the standard deviation, we can say we are 95% confident its accurate.
web survey
the invitation of the respondents and the completion of the survey over the internet or by email, university & college students love online surveys, design issues, 7-10 day turn around.
central limit theorem
the larger the sample size. 1. the more the pattern of the samples resembles a normal bell curve 2. the more the midpoint of the curve converges with the population parameter 3. the smaller the standard error (ex the more tightly various samples will cluster around the true population parameter).
prolonged engagement
the longer you stay in the field or the more people you interview can et you to go to truth more
operational
the measures, how are we going to measure this concept, what technique
independent variable
the presumed cause
dependent variable
the presumed effect
conceptualization
the process of coming up with definition
confidence interval
the range of values, usually a little higher and lower than a specified sample statistic, within which the true population parameter should fall
sampling ratio
the sample size in relation to are target population ex. 100/1000=10%
test re-test reliability
the whole idea of if you give someone a test tomorrow and again next week, they should be similar. ex. we shouldn't hate our job today and next week we absolutely love it
social groups
there can be strange dynamics that arise that do not happen in single interviews. ex some people might get angry at other, one person might dominate, etc.
passive smoke
they have a monetary incentive to deceive people and they didn't have complete or reliable data
open expression encouraged
they might not have said something before and then someone says something that reminds them of something that probes their memory and they can say "oh I have dealt with something like that" and then share their experiences, thoughts, etc.
push polling
they say something outrageous to try to have you change your vote.
indirect questions
those that the interviewer asks in order to get a sense of how the interviewee believes other people think. ex. do other students feel the same way about the teacher as you do?
laboratory settings
traditional. used in natural sciences and psychology
thick decription
transferability; you detail every step of research process from start to finish so if people look at these description
Linkert style questions
used for measuring simple attitudes, beliefs, emotions or behaviours. ex. when you think back to your high school years, you feel. 1. very unhappy, 2. somewhat unhappy, 3. don't feel anything, 4. happy, 5. very happy.
principles of question-writing
1. avoid jargon, slang or abbreviations 2. avoid ambiguity, confusion or vagueness 3. avoid emotional language and "loaded" words 4. watch out for prestige bias in your questions. 5. avoid double-barreled questions 6. do not confuse beliefs with reality 7. avoid leading questions 8. avoid questions beyond respondents' capabilities 9. avoid false premises or assumptions 10. if possible, do not ask respondent about future intentions. the link between intentions and future behaviour is tenuous. 11. try not to use negatives and double negatives in your questions 12. avoid overlapping or unbalanced response categories.
improving reliability
1. clearly conceptualize all constructs 2. increase the level of measurement (better rto measure at highest level of measurement (ex how often do you use yout cellphone in minutes in a day)- you're reying to take away any ambiguity. 3. use multiple indicators of a variable (someone overall might like their job, but there might be certain parts about it that they hate, you need to probe all dimnesions, have multiple indicators 4. use pretests, pilot surveys and replication (if you;re making a survey and going to give it to people, before you give it out, you give it to a small number of people or people who deal with the people you'll be giving it to, to make sure that its appropriate for the group, you redo it and take things out, etc.
informants
1. very familiar with culture or group we are trying to get into and in position to witness significant events 2. individual is currently involved in the culture 3. the person can spend time with the researcher 4. nonanalytical
bobo doll study
Albert Bandura. "transmission of aggression through imitation of aggressive models".
james patrick
Glasglow gang observed, semi participant; was a high school teacher, he knew there were a lot of violent teen gangs and asked one of his students if he could covertly enter the gang and he was brought in for 4 months with this gang until he could not do anymore but he was only semi-participant because he did not want to do all of the violent stuff that the rest were doing- people became suspicious
transferability
aka generalizability, because we have much smaller sample sizes, they question how we can generalize these responses to the general public/population
open ended questions
allows for free responses (qualitative analysis). its very hard to code an essay, you actually can't and it all represents different people so its only used for qualitative questions.
Silence
allows interviewee to reflect on what has been discussed (can encourage more talking).
specifying questions
allows interviewer to gain further information about a particular aspect of the interviewee's answer ex. can you elaborate on the event with that teacher who made you so upset?
cluster sampling
also called multi-stage cluster sampling. addresses two problems, lack of a good sampling frame for dispersed population, the cost of reaching a sampled element is very high, a "cluster" is a unit that contains final sampling units but is treated temporarily as a unit in itself, sample from larger clusters then within the clusters themselves. typically only used in nation side studies (ex stats canada uses these). ex. ill have the geographical cluster of every province, and then in each province ill pick a random sample of schools and then within those schools ill pick a certain number of teachers and that will give me my results
snowball sampling
also called network, chain referral or reputational sampling ex. you only need to know one homeless person and you ask them if they know any other homeless people and you ask them to get in contact with them, and they call you, and then their friends contact you, etc.
criterion validity
are we referring to some outside independent source for verification. ex. survey on depression and gave it out to people, teh result should be consistent with those with the established scale that has been around for a long time- if I am depressed on one scale, i should be depressed on the other. is the measure concurrent with other established measures?
transparency
are you being clear about methods and analysis? in surveys and such, we can show these and everything what we did, but because in qualitative our methods and guides are so different, we have so many differences within interviews, we can't say we have these clear cut methods and questions, etc. we can't be that transparent/
disclosure
are you giving the group the full knowledge of your role or are you adopting the role and not telling them you are a researcher
erving goffman
asylums, full participant; thanks to his research, he got asylum removed, asylums were like prison like places where they were just thrown in and all of their rights were removed, we went in as an employee and recorded everything and compared them to prisons and helped them get removed. ex. compared nurses to prison guards
probability sampling
based on theories of probability from mathematics, principle of randomization or chance, randomly selecting units from larger population allows us to provide inferences to the larger population (we can generalize)
introducing questions
beginning of interview; warming people up, getting the ball rolling
survey research
best method available to collect original information about a population, survey research is very old, very popular, research technique, 1880 survey by Karl Marx in France. topics in this: behaviours, attitudes/beliefs/opinions, characteristics, expectations, self classifications and knowledge
interval
everything that nominal and ordinal do PLUS can specify the amount of distance between categories. ex. temperature, IQ scores, dress size
favours
ex observing a daycare, one of the staff needs to take a break so you say "do not worry, I got this"
richness of data
ex. do not go to Tim Horton's to observe a gang and their criminal activity, you won't get a lot of data
friendly conversation
causual or friendly greeting initiates, no explicit goal, avoidance of repetition, expressions of interest and ignorance can be balanced, turn taking; encounter is balanced, pausing or silences acceptable.
survey
clear beginning and end, same questions, same order and same way, interviewer neutral, interview asks questions, respondent answers, professional tone, business like, close-ended questions dominant, attempt to standardize the experience.
credibility
confidence in the "truth" of the findings
member checking
controversial; only done with overt status, before you publish the results and you go back to the people and ask them if they have really captured what they meant to say- people might not like the way that they were captures what they meant to say- people might not like the way that they were captured and do not like what they said or "I did not say that to you or "take my data out"
trustworthiness
credibility, transferability, dependability and confirmability
diffusion of treatment
ex. experiment on memory technique, taught to one group and then let control and experiment groups hang out and they talked about it so now everyone knows and will use the technique
logic of PS
ex. we want to know the gender breakdown of all UOIT students, do we have time or money to track down and survey every single student? no!, we sample, but random chance means our sample may be off.. what to do? and draw many random samples and plot their distributions. we might randomly select a sample that does not reflect the population. sample size is key
experimental research
excellent for hypothesis testing (positivist), for explanatory research due to added control of experimental design and involves taking "action" and observing consequences of this action.
reliability
dependability or consistency. the same measure repeated under identical or very similar conditions should yield same results. how stable is this measure over time. ex if i give you a test tomorrow and then 2 weeks later, they should be similar
inquiry audit
dependability; bringing in another researcher and showing them the thick description and essentially audit you research, analyze and question everything from the same, methods, your biases,etc. and can give you a better objective ability to say it will be consistent if replicated
level of involvement
detached outsider at one extreme, intimately involved insider at the other and semi-participant. are you going to watch or secretly follow them around or fully participate. ex. if i go to the amazon to observe a tribe, am I going to fully immerse myself and be apart of the tribe and fully commit, or are you going to sort of commit and participate little or are you just going to stand back and watch?
qualitative measurement
devleop ways to capture and express concepts using various alternatives to numbers and inductive approach measures features of social life as part of a process that integrates concept and theory development with measurement. more iterative, can define and moderate measure while in data collection
history
did some unanticipated external event occur that could impact the dependent variable.
maturation
did the change in dependent variable happen because of some natural process. ex. during experiment on looing at kids play, want to see how they play in grades 1-6, kids in kindergarden play differently than kids in grade 6 because they change, develop, etc.
presentation of self
do you blend in or are you clearly an outsider and are making people uncomfortable (can cause reactivity)- have to make people feel comfortable so they act naturally around you and behave as if you were not there
standard form
don't get the option of no response. ex. here is the question about another country. do you agree or disagree with the statement? Russian leaders are basically trying to get along with america?
nonanalytical
don't want someone who is trying to theorize for you- just want an average member who will give you the inside information but won't be thinking too much an guide you to think about some things
questionnaire construction
each question represents a variable, usually translates into a variety of demographic questions and one or more scales and indexes. use existing scales or measures where possible, modify if necessary to meet the needs of your particular study. length is around 5 minutes, approximately 15 questions.
ratio
essentially the same as interval but has "true zero". ex. age, income (you can have the absence of money, but you can't have an absence of temperature when its at zero) etc. the absence of something is what we mean by zero. we always want to measure at his level when possible, its best to have this so if we want to simplify it down, we can, we can generalize it but if we don't have data at this level, we can't do that.
composition/number of groups
heterogeneous (different people) & homogeneous (same type of people). its better to have homogeneous people so they are able to build off of each other's ideas, and if there is too much variation you might not be able to relate or feel more judge and it you're the same like goes with like and feel more comfortable, you have to have multiple focus groups with different homogeneous groups depending on the diversity in the respondents you want, will have to have more depending on how many traits you want to account for
sampling error
how far off were our statistics from the actual parameter/ the actual population
internal consistency
how much do our measures together reflect this external measure. you want all of your internal measures to be consistent and to correlate. if you find question that does not correlate with the others you might need to get rid of it.
predictive validity
how we perform on one measure should indicate a future behaviour or outcome. ex. how well someone does on the MCATs or LSATs should indicate how well they go in either medical or law school
intrumentation
if using a particular instrument to measure change sin dependent ex. scale to measure weight loss and the scale recalibrates during the test so you have to constantly check to see if it still works
appearance of interest
if you are watching things you consider boring, you have to pretend everything is SO interesting, got to fake it
full- filter
if you don't have an opinion that's fine, but if you do, do you agree or disagree. ex. here is a question about another country. not everyone has an opinion on this. if you do not have an opinion, just say so. here is the statement: "russian leaders are basically trying to get along with america?" do you have an opinion on that? if yes, do you agree or disagree.
bandwagon phenomenon
if you see everyone if voting for one person, then you think you'll vote for them
treatment
independent variable. the stimulus or the manipulation.
ordinal measures
indicate a difference PLUS the categories can be ordered or ranked. ex. grades, "agreement scales" (ex. Linkert), frequency of doing something if asked in categories. different but we can order them, there is some ranking. likert scale (ex. strongly disagree, disagree, sort of disagree)
nominal measure
indicate different categories (ex. ethnicity, gender, religious affiliation, hair colour, type of car). the lowest level of measurement
matching
individuals are matched on important characteristics.
continuous
infinite number of values or attributes (ex. temp, age, income, crime rate, etc)
computer assisted telephone interviewing
interviewers sit at a computer terminal with a screen, answers directly entered into the computer, very fast and often used in polling and telemarketing.
structuring questions
interviewing keeping the interview on track ex. okay that was great, now can I ask you about this area?
kvale type questions
introducing, follow up, probing, specifying, direct, indirect, structuring, interpreting and silence
transferability
involved with trustwothiness; showing that the findings have applicability in other contexts
qualitative interviewing
involves asking questions, listening, expressing interest and recording EVERYTHING what was said, it is a joint production of a researcher and an interviewee, interviewee are active participants whose insights, feeling and cooperation are essential parts of a discussion process that reveal subjective meanings. study participants express themselves in the forms in which they normally speak, think and organize reality, a researcher retains members jokes and narrative stories, everything in their natural form and does not repackage them into a standardized format.
measurement validity
is something measuring what it is supposed to be measuring? face, content, criterion (concurrent & predictive). ex. if you ask question that don;t measure what I am trying to measure, the data will be invalid.
content validity
is the whole content of our measure accurately getting at what our conceptualized definition? are we missing certain dimensions? ex. all of the questions is the survey accurately reflect all of the dimensions in our definition
reflexivity
just look back and reflex on your experience and see if you were bias or objective and did things properly
groupthink
just think like the group, you are more inclined to think like the group and more likely to lean towards the group bias, want to avoid conflict by disagreeing, you might say something in the group but you do not actually feel that way
interrater reliability
level of agreement among different raters. ex when you want to look at things such as childhood aggression and you have a checklist to refer to when observing and you have to check it off when you see it, and the other rater said they saw 10 instances of aggression and the other only saw 3, you have to reevaluate your measures (ex "acts aggressive"- leaves a lot of it up to interpretation). ex. do you use your survey a lot of a little- what's a lot? what's a little? there is a lot of vagueness
sampling frame
list of everyone in that population and we number them
negative case analysis
looking specifically at cases that deviant from the dominant pattern
interpreting questions
making sure you interpret what they say correctly ex. now i just want to be sure, you feel your teacher treated you differently today because you dressed a certain way?
post test
measurement of dependent variable after treatment
pre test
measurement of independent variable before treatment, you test people on certain indicator. ex. see on the base level how aggressive they are.
random assignment
method for assigning cases (usually individuals) to groups for the purpose of making comparisons. it is a way to divide or sort a collection of cases into two or more groups in order to increase one's confidence that the groups do not differ in a systematic way.
leaving the field
more complicated than you might think! do I just leave right away, or tell people in advance, draw it out after a month or a few months. have to figure out how to tell people, what are you going to say, how much info you will give, etc.
quantitative measurement
much more concerned about details surrounding measurement, measurement protocols determined well before actual data collection and deductive approach begins with concept, moves to developing precise measures to capture that concept in numerical form. more precise.
population
name we give to larger group from which we draw our sample
field experiment
natural/real world setting, reduces reactivity found in classical experimental designs, tends to have higher EV and lower IX. (more generalizable but less controlled). subjects are usually unaware they are part of an experiment.
periodicty
no inherent ordering in that list (sampling frame)
public opinion polling
no official or mandatory regulation across Canada, bandwagon phenomenon vs free will effect. problems and discrepancies
non probability sampling
non-random samples, researchers have limited knowledge about the larger group or population from which the sample is taken, ex. no sampling frame exists. typically qualitative studies. ex. homeless people, people with depression in Canada, etc. these don;t have lists of all of the names of all of the people
attitude of strangeness
notice every ordinary detail and look as if it's the eyes of the stranger (everything is new and exciting), document everything
field researcher job
observes ordinary events & everyday activities as they happen in natural setting, becomes directly involved with the people being studied, personally experiences the process of daily social life in the field setting and acquires an insider's point of view while maintaining the analytic perspective or distance of an outsider. there is a fine line between getting immersed and going native, produces data in the forms of extensive written notes, as well as diagrams, maps, or pics to provide very detailed descriptions, notices both explicit & tacit aspects of culture, copes with a high level of personal stress, uncertainty, ethical dilemmas and ambiguity and sees events "holistically"
interviewer bias
occurs when the presence of the interviewer influences answers, an interviewers visible characteristics (race, gender) often affect interviews and respondent answers, a poorly trainer interviewer can affect respondent answers. their reactions to their responses might influence the way that they react. ex. asking about spousal abuse, asks a question like "have they hit you" and they answer yes and the interviewer goes "oh no,really?" or makes face, etc. then they could lie on following questions denying that they did anything
direct questions
often used later in the interview-questions that require quite brief answers, often yes/no. don't want a lot of these in qualitative
recoding
only if respondent says they do not want to be recorded
selective transcription
only taking away key points from the interview, not writing everything, its not preferable
subjects
people participating in experiment
haphazard, accidental convience
person on the street interview conducted by television programs, website polls. ex walk around the street and ask the first 10 people you see.
strategy for entering
planning (gatekeepers), negotiation and disclosure (overt vs covert)
telephone survey
popular method because about 95% of population can be reached by telephone, an interviewer calls a respondent, usually at home, asks questions and records answers (using interview schedule), samples drawn from: telephone directories, random digit dialing & non random samples), selecting a respondent from a household.
relations in the field
presentation of self, researcher as instrument, attitude of strangeness, and building rapport. normalizing social research and maintaining relations (exchange relationships aka favours, conflicts and appearance of interest)
probing questions
probes content of interviewee's answers but without giving away which parts of the answer are to be taken into account ex. can you elaborate on that?
statistical regression
problem of extreme values that can relate to both pretest and testing instrument and testing subjects. ex. redoing bobo with adults to see if we model behaviour of others, if we do this with convicted criminals, they will probably test as less violent in the retest because they are already so incredibly violent, if you test really bad at first, will go up, if start really high, will go down.
selection bias
problem when you don't use random assignment. ex. maybe let people themselves into 2 groups), they are automatically different and have this.
focus groups
procedure, the moderator, composition, number of groups and social groups
experimenter expectancy
reactivity in which experimenter itself inadvertently causes people to start to behave differently, not treatment impacting measure of dependent variable its subconscious efforts to change it. ex. medical drug trials, where they have drug and placebo, might be enthusiastic when giving people the real pill and then really unconcerned when giving the placebo pill),
control group
receives nothing at all
closed ended questions
refers to those questions which have fixed categories for answers (uses quantitative analysis). these are preferred.
simple random sampling
researcher develops accurate sampling frame, number each element (person) in the sample, selects elements from sampling frame in a mathematically random procedure, SRS with replacement or without replacement?, and locates those elements included in sample for study. we use software to create a table of random numbers and based on these numbers are the people we chose to use in the study.
stratified sampling
researcher first divides population into subpopulations (strata) on the basis of supplementary information, after the population is in strata, the researcher uses random sampling within the strata and this method guarantees representativesness of different strata within a sample. we still use systematic or random sampling, but first we separate by important strata. ex. if we want the sample to reflect an equal number of males and females, then we break them up by gender and then in those strata we do random or systematic sampling
dependability
showing that the findings are consistent and could be repeated
sequential sampling
similar to purposive except the researcher tries to find as many relevant cases until the amount of new information or diversity of cases is filled. theoretical sampling until we reach theoretical saturation. no number when you stop, try to reach saturation, try to find a new of new cases, but you only stop when you're not learning anything new, you're reach analytical saturation.
systematic sampling
simple random sampling with a shortcut, number each element of the sampling frame, instead of calculating a list of random numbers, calculate a sampling interval (every kth element) and use a random start between 1-10. helpful way to determine a sampling interval- population size over sample size.
sampling element
single case from that population we are sampling ex. an individual student
quota sampling
slight improvement over haphazard. stages: first identifies relevant categories of people (male & female, under 25, 25-40, over 40), then decides how many to get in each category. used in marketing sampling, have to get a certain number of people who fall into different categories.
population parameter
some sort of characteristic based on that population (ex 50% of UOIT students are male and other 50% is female)
devil's advocate
sometimes how we can avoid think, we might plant a participant or tell someone to disagree against the groups opinion, can help avoid group think by making people more comfortable to express a divergent opinion
snowball
starts with one person and then you network to other people in the same group ex. if you know other people in this group, can you give them my info. and tell them to give me a call and ill interview them
validity
suggest truthfulness and refers to the match between a construct and the way a researcher conceptualizes an idea or conceptual definition and a measure. am I getting at the truth of what this concept is? ex. if my measures don't actually give me results on what I am trying to study, the data is useless
role of interviewer
survey interview-special kind of social relationship, many respondent unaware of interviewers role (ex intimate chat/therapy session), good interviewers: obtain cooperation, build rapport, non-judgmental (poker face), highly structured interview protocols, and 40% of interviewers time spent on interviewing.
fafce to face interview
surveys conducted in person, using an interview schedule with answers recorded by the interviewers, greater success asking a wider range of questions (ex complex, contingency and open-ended questions), lenghtier survey instruments=training crucial
land humphreys
tearoom trade (participant-observation)
purposive sampling
used in situation when an "expert" uses judgement in selecting cases with a specific purpose in mind. appropriate when: using it to select case that are especially informative, using it to select members of hard to reach, specialized populations and using it to identify particular types of cases for in-depth investigation. ex. you go to people who put you in contact with people you want to study
follow up questions
used to extend the interviewee's answers to previous questions. ex. today I had a bad experience with so and so and then you ask oh what happened?
deviant case sampling
used when a researcher wants to study cases that differ from the "dominant pattern", cases are selected because they are unusually. ex. heroine addicts who are still really successful-how can they still be successful?
matrix questions
useful when you have a number of related items. composed of a series of questions that have identical response categories which are presented in table format with the response category labels at the top of the chart.
triangulation
using multiple methods (ex going to observe the people and then interview a few people from the group
full transcription
very time consuming, ex. 1 hour speech, takes 6 hours to transcribe). we include every single works as well as non verbal communication. ex. if people have a very visible expression or awkwardly laughs or is fidgeting.
the marshmellow experiment
walter mischel. kid was at table, asked what candy they liked. comes back and has 2 of their candies, said you can have one now and then I am going outside for 15 mins and if you wait you can have 2, they did follow up 30 years later, the kids who waited and could delay gratification did so much better in a lot of aspects of life than those who could not wait.
collecting field data
watching & listening (taking notes)
hidden populations
we are looking for people in which there is no list of all of the names exists, we don't have sampling frames.
researcher as instrument
we have to be aware that we have a lot of subjective feelings/reactions that could influence our interpretations
naturalism
we want to study people in their true real life social events, see how they occur in everyday life
Kenneth Good
went to a tribe to study them for 12 years, married a 13 year old girl and has kids and took her back to his home, there was huge culture shock and she left back to her tribe
mortality
when people literally die
testing
when the pretest measure itself effects the experiment ex. give exam on 1st day of class, get results and then give exact same test on last day and your theory is how good of a teacher you are, but the teaching might not be better, it is just because you paid attention to pretest so did better on post test)
unfamiliarity
when you go to a place 100 times, you can miss details, when you walk into a place you have never been you spend more time looking at things you've never seen, you can see more unfamiliar things, at common places everything blends in
field site
where the observation occurs
access
without a gatekeeper, they might not get access into the culture, the gatekeeper negotiates their access in (either overtly or covertly), most likely a member of the group, but even when in there, they have to gain trust of every member, public spaces are easier than private spaces
deception/debriefing
you aren't telling people true hypothesis, and when this happens you have to tell people right away what really happened
polarization effect
you go into focus group and you do not have strong opinions on topic and then you spend time there and then you leave and have these strong opinions on it, and then if you perform the same focus group the next week you will probably get totally different results
negotiation
you have to continually negotiate with the members of the group to have them continue to accept you
planning
you need to bargain with people who are a part of the group you are trying to look at (ex a member of a gang, a tribe chief, etc)
free will efefct
you see something happening and then you may want to change it
conflicts
you should NOT get involved, could make people feel alienated and then they won't want to cooperate anymore
suitability
your ascripted characteristics (ex race, gender, age, etc) might make you unsuitable character to go into the field site